Elevator Acceleration Sensing: Design and Estimation Recognition Algorithm Using Crowdsourcing

In human activity sensing areas, we focus on elevator recognition using accelerometer. In this paper, our main challenge is to create an elevator acceleration recognition algorithm. In order to put this algorithm into real world use, This algorithm should be robust enough that can ignore the differences of the elevators, devices and subjects. To realize this aim, we build a crowdsourcing system which consists of an android application and a server. Android users can use this application to calculate elevator's moving distance. They can also upload elevators' accelerometer signals to the server, so we can get elevator acceleration data from all over the world. after publishing the system, we use the data we collected to build an elevator acceleration database. By using this database, we build a robust elevator recognition algorithm. we also use the features of elevators to distinguish them.

I. Introduction

Recently, with the development of MEMS technology, the smartphone which has many kinds of sensors are more and more popular. There are many kinds of sensors, such as microphone, camera, gyro, accelerometer and etc. we can get the sensor signals by using these sensors. The users always carry these sensors in their daily life, if we record signals from the sensors, we can get user's many information from the sensors. By using these information, we can build life management service, life log, navigation system, or context aware services.